Matrix Metalloproteinases of Epithelial Origin in Facial Sebum of Patients with Acne and their ...

12 Pages·2005·0.36 MB·English

Matrix Metalloproteinases of Epithelial Origin in Facial Sebum of
Patients with Acne and their Regulation by Isotretinoin
Eleni Papakonstantinou,(cid:1)1 Alexios J. Aletras,w1 Evelyn Glass,z1 Panagiotis Tsogas,(cid:1)1 Alexander
Dionyssopoulos,y James Adjaye,z Sabine Fimmel,z Panagiotis Gouvousis,w Ralf Herwig,z Hans Lehrach,z
Christos C. Zouboulis,2z and George Karakiulakis(cid:1)2
(cid:1)DepartmentofPharmacology,SchoolofMedicine,AristotleUniversity,Thessaloniki,Greece;wLaboratoryofBiochemistry,DepartmentofChemistry,Universityof
Patras,Patras,Greece;zDepartmentofDermatology,Charite´ UniversityMedicineBerlin,CampusBenjaminFranklin,Berlin,Germany;yDivisionofSkinOncologic
Surgery-PlasticSurgery,SchoolofMedicine,AristotleUniversity,Thessaloniki,Greece;zDepartmentofVertebrateGenomics,MaxPlanckInstituteforMolecular
Genetics,Berlin,Germany
Acnevulgarisisaskindisorderofthesebaceousfollicles,involvinghyperkeratinizationandperifollicularinflam-
mation. Matrix metalloproteinases (MMP) have a predominant role in inflammatory matrix remodeling and hyper-
proliferative skin disorders. We investigated the expression of MMP and tissue inhibitors of MMP (TIMP) in facial
sebumspecimensfromacnepatients,beforeandaftertreatmentwithisotretinoin.GelatinzymographyandWest-
ern-blot analysis revealed that sebum contains proMMP-9, whichwas decreasedfollowing per os or topical treat-
ment with isotretinoin and in parallel to the clinical improvement of acne. Sebum also contains MMP-1, MMP-13,
TIMP-1,andTIMP-2,asassessedbyELISAandwesternblot,butonlyMMP-13wasdecreasedfollowingtreatment
with isotretinoin. The origin of MMP and TIMP in sebum is attributed to keratinocytes and sebocytes, since we
foundthatHaCaTkeratinocytesinculturesecreteproMMP-2,proMMP-9,MMP-1,MMP-13,TIMP-1,andTIMP-2.SZ95
sebocytes in culture secreted proMMP-2 and proMMP-9, which was also confirmed by microarray analysis. Is-
otretinoininhibitedthearachidonicacid-inducedsecretionandmRNAexpressionofproMMP-2and-9inbothcell
types and of MMP-13 in HaCaT keratinocytes. These data indicate that MMP and TIMP of epithelial origin may be
involved in acne pathogenesis, and that isotretinoin-induced reduction in MMP-9 and -13 may contribute to the
therapeutic effects of the agent in acne.
Key words: acne/isotretinoin/keratinocytes/MMP/sebocytes/TIMP
J Invest Dermatol, 2005
Acne vulgaris is the most common skin disorder, initiated topical treatment of hyperproliferative skin disorders, in-
due to sebaceous gland hyperactivity and hyperseborrhea, flammatory diseases, and cancer (Orfanos et al, 1997;
abnormal hyperproliferation of ductal keratinocytes and Zouboulis, 2001c), the non-aromatic retinoid isotretinoin
keratinization of the acroinfundibular epithelium of the se- (13-cis-RA) secures effective treatment of acne, but it may
baceous follicle, and inflammatory signaling giving rise to also cause adverse and toxic effects (Zouboulis and Orfa-
microcomedones(Zouboulis,2001a).Sebumconsistsprin- nos, 2000). The antiproliferative effectof isotretinoin on se-
cipally of triglycerides, which are subsequently hydrolyzed bocytes is manifested through its isomerization into all-
toyieldfreefattyacidsandglycerol,andinadditionsebum trans-RA and binding to nuclear RA receptors (Zouboulis,
also contains keratinocytes, microorganisms, neutrophils, 2001b; Tsukada et al, 2002). Isotretinoin displays key reg-
andmacrophages(ToyodaandMorohashi,2001).Although ulatory functions on epidermal growth and differentiation
individual acne lesions spontaneously regress, persistent (Fisher and Voorhees, 1996) but the cellular and biochem-
cases of acne vulgaris often require pharmacological inter- ical alterations associated with them are not fully clarified.
vention. Isotretinoin has been reported to affect matrix metal-
Pharmacotherapy of acne vulgaris includes a variety of loproteinases (MMP) (Jimenez et al, 2001; Zhu et al, 2001;
compounds, with the retinoids possessing a prevailing po- Devy et al, 2002), a family of Zn-dependent metal-
sition (Zouboulis, 2001c). Among the retinoic acid (RA) an- lopeptidases that has been implicated in skin biology dur-
alogues that have been established for systemic and/or ing inflammatory matrix remodeling, neovascularization,
wound healing, and malignant transformation. Thus, it ap-
pears that MMP have a predominant role in pathological
Abbreviations:aRNA,amplifiedRNA;cDNA,complementaryDNA; manifestations of diseases treated with retinoids, such as
MMP, matrix metalloproteinases; RA, retinoic acid; SDS-PAGE, hyperproliferative skin disorders, inflammatory diseases,
sodiumdodecylsulfate-polyacrylamide gel electrophoresis;TIMP,
and cancer (Orfanos et al, 1997). MMP degrade extracel-
tissueinhibitorsofmetalloproteinases
1Thefirst4authorshavecontributedequallytothismanuscript. lularmatrixmoleculesduringphysiologicalandpathological
2Thelast2authorsshareseniorauthorship. tissueremodeling(VisseandNagase,2003).TheMMPgene
Copyrightr2005by TheSocietyforInvestigativeDermatology,Inc.
1
2 PAPAKONSTANTINOU ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
family encodes for several proenzymes (proMMP) with revealed the presence of P. acnes (propionibacterium
common and distinctive structural and functional proper- acnes) in all 0 d samples and of Staphylococcus epider-
ties,classifiedas:(a)collagenases1–4(MMP-1,-8,-13and midis in two samples. No other microorganisms were de-
-18, respectively), (b) gelatinases A and B (MMP-2 and -9, tected in 0 d samples. Furthermore, no bacteria could be
respectively), (c) stromelysins 1–3 (MMP-3, -10, and -11, detected in any of the 60 d samples.
respectively), (d) matrilysins 1–2 (MMP-7 and -26, respec-
tively),(e)membranetype1–6(MT-MMP:MMP-14,-15,-16, Gelatinase activity Gelatin zymography analysis revealed
-17, -24, and -25, respectively), and (f) various others that0d sebumsamplesoffacial acnelesionsexpressgel-
(MMP-12, -19, -20, -21, -23, -27, and -28) (Visse and atinase activity, which produced major lysisbands (Fig1a),
Nagase,2003).TheactivityofMMPistightlyregulateddur- which comigrated as purified proMMP-9 (92.0 kDa). Gela-
ingproMMPexpression(Stetler-Stevensonetal,1993) and tinolyticactivitywascompletelyinhibitedbydevelopingthe
byproteolytic activation ofsecretedlatentMMP (Caoetal, zymogramsinthepresenceofthemetalchelatorsNa2EDTA
1995). MMP, both in latent and activated form, are also (20mM)or1,10-phenanthroline(4mM),butwasunaffected
regulatedbytheformationofstoichiometric1:1complexes byN-ethyl-maleimide(5mM)(datanotshown),indicativeof
with specific inhibitors, known as tissue inhibitors of met- a metalloproteinase, corresponding to proMMP-9. Zymo-
alloproteinases (TIMP) (Stetler-Stevenson et al, 1993). Four graphy analysis of samples obtained at 30 and 60 d fol-
TIMPhavebeenidentified,butinactivationofMMPbyTIMP lowingsystemicperosortopicaltreatmentwithisotretinoin
ispredominantlyattributedtoTIMP-1and-2.TIMP-1forms (Fig 1a), indicated that gelatinolytic activity was reduced
high-affinity, non-covalent complexes with latent MMP-9 in relation to the duration of treatment. Quantification of
andactiveMMP-1,-3,and-9,whereasTIMP-2formscom- zymography lysis bands of all acne samples tested, using
plexes with latent and active MMP-2 (Gomez et al, 1997). a computer-assisted image analysis program (Fig 1b), re-
TIMP-1/proMMP-9 and TIMP-2/proMMP-2 complexes are vealed that thedecreasein gelatinolytic activity was statis-
also capable of inhibiting active MMP through the unoccu-
piedN-terminaldomainoftheinhibitorinthecomplex(Itoh
et al, 1995).
TheeffectofisotretinoinonMMP,however,iscontradic-
tory, since there is evidence suggesting that retinoids in-
crease (Jimenez et al, 2001), reduce (Neuville et al, 1999;
Leville et al, 2000; Axel et al, 2001; Frankenberger et al,
2001; Tsang and Crowe, 2001; Devy et al, 2002; Osteen
etal, 2002),ordonot affect(Zhuet al, 2001) theactivity or
gene expression of MMP in various human and animal bi-
ological systems and cell lines. Furthermore, there is no
report concerning the presence of MMP in lesions of acne
vulgaris or the effect of isotretinoin treatment on MMP in
sebum of acne patients.
In this study, we investigated the presence of gelatin-
ases, collagenases, and TIMP in sebum samples of facial
lesions from patients with acne vulgaris, as well as in cul-
turesofHaCaTkeratinocytesandSZ95sebocytes,andthe
effect of isotretinoin in these parameters. We found that
proMMP-9, MMP-1, MMP-13, and TIMP-1 and TIMP-2 are
present in the sebum of facial lesions from acne patients,
possiblyoriginating fromkeratinocytes andsebocytes, and
that MMP-9 and -13 are reduced in parallel to treatment
withisotretinoinandtheclinicalimprovementofthelesions.
Figure1
Results
Gelatinaseactivityisreducedinsebumoffacialacnelesionsina
time-dependent mannerduring therapy with isotretinoin. (a) Rep-
Acne sebum samples resentative gelatin zymography in aliquots of sebum (5 mg of protein)
duringsystemicperos(lanes4–6)andtopical(lanes7–9)treatmentwith
isotretinoin. Arrows indicated pre-stained standard protein molecular
Keratinocyte and bacteriologic findings Examination of se- weight markers (lane 1): phosphorylase b (97.4 kDa), bovine serum
bum specimens, diluted to contain the same amount of albumin (66.2 kDa), L-glutamic dehydrogenase (55.0 kDa), and oval-
protein per mL, revealed the presence of keratinocytes un- bumin (42.7 kDa). Lane 2: migration of purified promatrix metal-
loproteinase(proMMP)-2(72kDa);lane3:proMMP-9(92kDa);lanes4
der a light microscope in all 0 d samples from 59 patients.
and7:0dsamples;lanes5and8:30dsamples;lanes6and9:60d
After 60 d of topical (0.05% once daily) or systemic (1 mg samples.(b)Quantitativeanalysisofgelatinolyticactivityinsebumfol-
perkgperdperos)treatmentwithisotretinoin,examination lowing systemic per os (n¼23) or topical (n¼36) treatment of acne
of 60 d sebum samples showed that keratinocytes were patients with isotretinoin using a computer-supported image analysis
program.Eachbarrepresentsthemean(cid:1)SDfromnspecimens.Sta-
present only in 11 of 59 patients and in considerably di- (cid:1) (cid:1)(cid:1)(cid:1)
tisticalsignificance: po0.05; po0.01ascomparedwith0dsam-
minished numbers. Bacterial cultures of sebum specimens ples,whichwereregardedas100%gelatinolyticactivity.
2005 MATRIX METALLOPROTEINASESIN ACNEVULGARIS 3
tically significant at 30 d (po0.05) and 60 d (po0.01) of described bands of MMP-9 (Fig 2a, lane 5, per os treat-
treatment, wherreas the route of administration of is- ment). Quantification of the chemiluminescence of each
otretinoin did not influence the decrease in gelatinase band using a computer-assisted image analysis program
activity. revealed that the observed reduction was statistically sig-
nificant at 30 d of treatment (po0.01, po0.05, po0.01,
Detection of gelatinases and effect of isotretinoin treat- respectively, for each band) (Fig 2b), whereas the route of
ment Using human MMP-2 antiserum, we found that there administrationofisotretinoindidnotinfluencethedecrease
was no immunoreactivity for MMP-2 in 0 and 30 d sebum in MMP-9 immunoreactivity (results not shown).
samples (Fig 2a, lanes 2 and 3). Using human MMP-9 an-
tiserum, three distinct immunoreactive bands with M 94.0, Effect of isotretinoin on gelatinases in vitro In order to in-
r
78.0,and64.0kDawererevealedin0dsebumsamplesby vestigate whether isotretinoin has a direct inhibitory effect
western blotting (Fig 2a, lane 4). The upper band corre- ontheactivityofgelatinases,wesubjectedsebumsamples
sponds to proMMP-9 and comigrated as reference sample (0 d, corresponding to 5 mg protein) or commercially avail-
ofproMMP-9(94.0kDa)thathasbeenpreviouslyfullychar- able proMMP-2 and -9 (1 ng) to gel electrophoresis, re-
acterized from human periprosthetic tissue of loose hip moved sodium dodecyl sulfate (SDS), and conducted
endoprostheses (Syggelos et al, 2001). The two lower gelatin lysis of the gels by incubating in enzyme buffer in
immunoreactive bands may be attributed to endogenous the presence of isotretinoin (0–1 mM) for 18 h, at 371C.
activationordegradationofthe94.0kDaproenzyme,dueto Alternatively, we pre-incubated sebum samples (0 d, cor-
the experimental conditions adopted (Kerkela and Sa- respondingto5mgprotein)orcommerciallyavailableproM-
arialho-Kere, 2003). The fact that the 78.0 and 64.0 kDa MP-2 and -9 (1 ng), with isotretinoin (0–1 mM) for 30 min
proteins identified by western blotting could not be visual- at room temperature, subjected samples to gel elect-
izedbygelatinzymographyindicatesthattheseproteinsdo rophoresis, removed SDS, and conducted to gelatin lysis
notpossessgelatinolyticactivity.Treatmentwithisotretinoin ofthegelsbyincubatinginenzymebufferfor18h,at371C.
for 30 d resulted in reduced expression of all the above- Analysisoftheresultsrevealedthatneitherthepresenceof
isotretinoin during gelatin zymography (results not shown)
norpre-treatmentofthesampleswithisotretinoinfor30min
influenced proMMP-9 present in 0 d sebum samples (Fig
3a) or purified latent gelatinases (Fig 3b).
Detection of collagenases by western blotting and effect of
isotretinointreatment UsingpolyclonalantibodiesforMMP-
1,threedistinctimmunoreactivebandswererevealedin0d
sebumsamplesbywesternblotting(Fig4a,lane2),withM
r
52.4, 45.0, and 43.5 kDa as estimated using molecular
weightmarkers.Thesebandscorrespondedtopurifiedhu-
manMMPstandards:proMMP-1(52.4kDa)andfullyactive
MMP-1 (45.0 and 43.5 kDa), respectively, in comparison
with reference samples of proMMP-1 and MMP-1 (Fig 4a,
lane1),whichhavebeenpreviouslyfullycharacterizedfrom
human periprosthetic tissue of loose hip endoprostheses
(Syggelos et al, 2001). The ratio of latent to active MMP-1
wasapproximately8/1.Treatmentwithisotretinoinfor30or
60ddidnotaffecttheexpressionoflatentoractiveMMP-1
(Fig 4a, lane 3, topical treatment, 30 d).
Using polyclonal antibodies for MMP-13, two distinct
immunoreactivebandswererevealedin0dsebumsamples
Figure2 by western blotting (Fig 4b, lane 1), with M 65.0 and 52.0
r
Western blot analysis of gelatinases in sebum samplesindicates kDa as estimated using molecular weight markers. These
the presence of matrix metalloproteinase (MMP)-9 but not of
bands corresponded to purified human MMP standards:
MMP-2.(a)Representativewesternblotanalysisofsebumduringsys-
temicperostreatmentwithisotretinoin.Analysiswasperformedusing proMMP-13andfully active MMP-13,respectively, incom-
humanMMP-2antiserum:lane1:purifiedpromatrixmetalloproteinase parison with reference samples of proMMP-13 and MMP-
(proMMP)-2(72.0kDa),lane2,0dsample;lane3,30dsample,and
13 (Fig 4b, lane 3), which have been previously fully char-
human MMP-9 antiserum: lane 4, 0 d sample; lane 5, 30 d sample.
ArrowontherightindicatesthereferencesampleofproMMP-9(94.0 acterized from human periprosthetic tissue of loose hip
kDa),whichhasbeenpreviouslyfullycharacterizedfromhumanperi- endoprostheses(Syggelosetal,2001).Theratiooflatentto
prosthetic tissue of loose hip endoprostheses. Arrows on the left in- active MMP-13 was approximately 1/2. Treatment with is-
dicate molecular weight size markers (phosphorylase b, 97.4 kDa;
otretinoinfor30or60dsignificantlyreducedtheexpression
bovineserumalbumin,66.2kDa;L-glutamicdehydrogenase,55.0kDa).
(b) Quantitative analysis of western blots of MMP-9 immunoreactive oflatentoractiveMMP-13(Fig4b,lane3,topicaltreatment,
bandsofsebumfollowingsystemicperostreatmentofacnepatients 30 d).
(n¼23) with isotretinoin for 30 d, using a computer-supported image
analysis program. Each bar represents the mean(cid:1)SD from n spec-
(cid:1) (cid:1)(cid:1)(cid:1) Determination of collagenases by ELISA and effect of is-
imens.Statisticalsignificance: po0.05; po0.01ascomparedwith
0dsamples. otretinoin treatment Measurement of MMP-1 and -13 by
4 PAPAKONSTANTINOU ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure3
Isotretinoindoesnotinfluencegelatinaseinvitro.(a)Sebumsam-
ples (0 d, corresponding to 5 mg protein) were preincubated with is-
otretinoin(0–1mM)for30minatroomtemperatureandsubjectedto
Figure4
gelatinzymography,asdescribedinMaterialsandMethods.Lane4,0
Matrix metalloproteinase (MMP)-1 and MMP-13 are expressed in
mM;lane5,0.01mM;lane6,0.1mM;lane7,1mM.(b)Commercially
sebumoffacialacnelesions,andMMP-13expressionisreduced
available purified promatrix metalloproteinase (proMMP)-2 and proM-
during isotretinoin treatment. (a) Representative western blot anal-
MP-9(1ngprotein)weretreatedasabove.ProMMP-2:lane4,01mM;
ysisofsebumduringtopicaltreatmentwithisotretinoinusingpolyclonal
lane5,1mM.ProMMP-9:lane6,0.1mM;lane7,1mM,lane7.Arrows
antibodiesforMMP-1.Lane1andarrowsontheleft:immunoreactive
andlane1indicatemolecularweightmarkers:phosphorylaseb(97.4
bands from referencesamples of promatrix metalloproteinase (proM-
kDa), bovine serum albumin (66.2 kDa), L-glutamic dehydrogenase MP)-1 (52.4 kDa) and MMP-1 (45.0 and 43.5 kDa), which have been
(55.0kDa),ovalbumin(42.7kDa),andaldolase(40.0kDa).Lane2:pu-
fullycharacterizedpreviouslyfromhumanperiprosthetictissueofloose
rifiedproMMP-2(72kDa)andlane3purifiedproMMP-9(92kDa).
hip endoprostheses. Lane 2: immunoreactive bands of 0 d sebum
samples. Lane3:immunoreactive bandsof30 dsebumsamplesfol-
lowing topical treatment with isotretinoin. Lane 4 and arrows on the
rightindicateproteinmolecularweightmarkers:bovineserumalbumin
ELISA revealed that sebum samples expressed both colla-
(66.2 kDa), ovalbumin (42.7 kDa), and soybean trypsin ihibitor (21.5
genases(Table I,Fig 4c).In 0dsebum samplesof patients kDa).(b)Representativewesternblotanalysisofsebumduringtopical
groupedforperostreatment,MMP-1was105(cid:1)20pgper treatment with isotretinoin using polyclonal antibodies for MMP-13.
Lane1andarrowsontheleftindicateproteinmolecularweightmark-
mgofproteinandMMP-13was92(cid:1)8pgpermgofprotein.
ers:bovineserumalbumin(66.2kDa),ovalbumin(42.7kDa)andsoy-
Similar values were obtained in 0 d sebum samples from beantrypsinihibitor(21.5kDa).Lane2:immunoreactivebandsof0d
patientsgroupedfortopicaladministration.MMP-1wasnot sebumsamples.Lane3:immunoreactivebandsof30dsebumsam-
affected following treatment with isotretinoin for 60 d, irre- plesfollowingtopicaltreatmentwithisotretinoin.Lane4andarrowson
theright:immunoreactivebandsfromreferencesamplesofproMMP-
spectiveoftherouteofadministration.Incontrast,MMP-13
13(65.0kDa)andMMP-13(52.0kDa)fromhumanperiprosthetictis-
was significantly reduced by about 55% and 58% after 30 sue. (c) Aliquots of sebum samples were assessed for MMP-1 and
and 60 d of per os treatment with isotretinoin, respectively MMP-13 by ELISA, prior to and following per os (n¼23) and topical
(n¼36) treatment of acne patients with isotretinoin, for 30 and 60 d.
(po0.02). This effect was also evident after topical admin-
Eachbarrepresentsthemean(cid:1)SDoftriplicatedeterminationsfromn
istrationofisotretinoin,resultinginasignificantdecreaseof samples. Statistical significance: (cid:1)(cid:1)po0.02 as compared with 0 d
MMP-13byapproximately50%and60%after30and60d sample.
of treatment with isotretinoin, respectively (po0.02).
HaCaTkeratinocyteandSZ95sebocytecultures Inorder
Production of TIMP Measurement of TIMP-1 and -2 by toelucidatetheoriginofMMPandTIMPinsebumsamples,
ELISAin0dsebumsamplesofpatientsgroupedforperos we examined their expression in HaCaT keratinocytes and
administrationindicatedthepresenceofTIMP-1(405(cid:1)198 SZ95sebocyteculturesafter12,24,and48hofincubation
pgpermgprotein)andTIMP-2(219(cid:1)84pgpermgprotein) withorwithoutarachidonicacid,inthepresenceandinthe
(Fig5).Similarvalueswereobtainedin0d sebumsamples absence of isotretinoin. Arachidonic acid, a pro-inflamma-
from patients grouped for topical administration. Treatment tory essential fatty acid, was included in the study since it
with isotretinoin did not influence the production of TIMP, hasalsobeenreportedtoinducesebaceouslipidsynthesis
irrespective of the route of administration (Fig 5). (Wro´bel et al, 2003) and may be involved in acne patho-
2005 MATRIX METALLOPROTEINASESIN ACNEVULGARIS 5
TableI. Expressionofcollagenases,gelatinases,andTIMPinacnesebum,HaCaTkeratinocytes,andSZ95sebocytesasdetected
bygelatinzymography,westernblotting,ELISA,andRT-PCR
HaCaTkeratinocytes SZ95sebocytes
Sebumofacne
patients:proteins Proteins mRNA Proteins mRNA
Nomenclature Untreated Iso Untreated AA Iso Untreated AA Iso Untreated AA Iso Untreated AA Iso
Collagenases
MMP-1 D $ D " $ D " $ ND ND ND ND ND ND
MMP-13 D # D $ #c D "c # ND ND ND ND ND ND
Gelatinases
MMP-2a ND ND D " #b D " #b,c D " #b D " #b,c
MMP-9a D # D " #b D " # D " #b D " #b
TIMP
TIMP-1 D $ D " $ D " $ ND ND ND ND ND ND
TIMP-2 D $ D $ $ — — — ND ND ND — — —
aProteinsand/orenzymeactivity.
bReductioninthearachidonicacid-treatedsamplesonly.
cNotstatisticallysignificant.
Iso,isotretinointreated;AA,arachidonic acidtreated;D,detected;ND,notdetected; $,notaffectedbytreatment; #,reducedbytreatment;",
increasedbytreatment;—,notexamined;TIMP,tissueinhibitorsofMMP,MMP,matrixmetalloproteinases.
genesis (Zouboulis et al, 2005). Cells cultured in the pres- replicate experiments indicated the detectability. The level
ence of arachidonic acid will better represent follicular ker- of 0.9 was used to judge expression of genes. For each
atinocytes of acne patients than HaCaT keratinocytes not cDNA’sexpressionsignal,thelogratio(base2)ofthissignal
under inflammatory challenge. was computed with the median expression signal of all
15,657cDNA onthe array. We foundthat while gelatinases
wereexpressed,mRNAforcollagenasesMMP-1and-13as
MMP and TIMP mRNA expression in SZ95 sebocytes The
well asfor TIMP-2was not detected (Table I). Furthermore,
mRNA expression of MMP and TIMP in SZ95 sebocytes
SZ95 sebocytes expressed with high probability (p40.95)
was investigated by analyzing the results of expression
mRNAforMMP-7,-14,-15,-21,and-24(eventuallyMMP-3
profiles generated from untreated SZ95 sebocytes, follow-
and-11,too)aswellasforTIMP-3and-4,whereastheydid
ing complementary DNA (cDNA) microarrays and image
notexpressmRNAforMMP-10,-12,-16,-19,-20,-25,-26,
analysis (Fig S1). Expression was judged by signal detect-
-27, and -28 (Fig S1). Caution should, however, be exer-
ability using a negative control sample present on each ar-
cisedininterpretingtheseresults,sincesomeoftheabove-
ray. The average proportion of negative samples that was
mentionedMMPandTIMP-2thatcouldnotbedetectedby
expressed below the probe’s signal threshold across the
cDNA microarrays may be expressed if they are examined
by more sensitive methods such as quantitative PCR.
Gelatinase activity Gelatin zymography analysis revealed
that both HaCaT keratinocytes and SZ95 sebocytes, after
24hinculture,secretedgelatinaseactivity,whichproduced
two major lysis bands (Fig 6a and c, respectively, Table I).
The upper band comigrated as commercially available
proMMP-9, with M corresponding to 87.0 kDa. The lower
r
band comigrated as commercially available proMMP-2,
withM correspondingto68.0kDa.InHaCaTkeratinocytes,
r
the minor lysis band with M 65.0 kDa, comigrating as
r
commerciallyavailableMMP-2,mayrepresentactivationof
proMMP-2. In SZ95 sebocytes, minor lysis bands with
Figure5
M487.0 kDa may be attributed to disulfide polymers of
Tissue inhibitor of matrix metalloproteinase (TIMP)-1 and -2 are r
expressed in sebum samples of facial acne lesions but are not MMP molecules. Gelatinolytic activity was completely in-
influenced during isotretinoin treatment. TIMP were measured by hibitedbydevelopingthezymogramsinthepresenceofthe
ELISAinaliquotsofsebumspecimens(2mgofprotein)duringperos metal chelators Na EDTA (20 mM) or 1,10-phenanthroline
treatmentof23acnepatientsandtopicaltreatmentof36acnepatients 2
(4 mM), but was unaffected by N-ethyl-maleimide (5 mM)
with isotretinoin for 0, 30, and 60 d. Each bar represents the
mean(cid:1)SDoftriplicatedeterminationsfromnspecimens. (Fig 6a and c), indicating that the activity is due to metal-
6 PAPAKONSTANTINOU ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
loproteinaseandexcludingserineorcysteinproteinaseac-
tivity, respectively. These results indicate that the 2 major
lysis bands described above are MMP, corresponding to
proMMP-9 and -2, respectively (Karakiulakis et al, 1988;
Visse and Nagase, 2003). Quantification of zymography
lysis bands obtained from all cell cultures tested revealed
that arachidonic acid induced the secretion of MMP-9 by
45% (po0.02) and of MMP-2 by 30% (po0.05) in HaCaT
keratinocytes, and by 300% (po0.01) and 25% (po0.05),
respectively, in SZ95 sebocytes. Zymography analysis of
Figure7
Matrix metalloproteinase (MMP)-1 is secreted by HaCaT keratin-
ocytesandupregulatedbyarachidonicacid.Aliquotsofsupernat-
antsfromHaCaTkeratinocytesincubatedfor12–48hwereassessed
forMMP-1byELISA.Eachbarrepresentsthemean(cid:1)SDoftriplicate
determinationsfromfourassays.Iso,isotretinoin,AA,arachidonicacid
(10(cid:2)4 M). Statistical significance: (cid:1)po0.05; (cid:1)(cid:1)po0.02, as compared
withcontrol(untreatedcells).
samples obtained at 12 or 48 h in culture revealed similar
profilesforbothcelllines(resultsnotshown),asfor24hin
culture.Isotretinoin(at10(cid:2)8and10(cid:2)7M)didnotaffectgel-
atinolyticactivityinkeratinocytesorSZ95sebocytesatany
time point (Fig 6, for 24 h in culture). In both cell lines,
however,isotretinoinreducedthearachidonicacid-induced
secretionofMMP-2and-9tolevelscomparabletocontrols
(po0.05 to 0.01) (Figs 6b and d).
Secretionofcollagenases MeasurementofMMP-1and-13
byELISArevealedthatHaCaT keratinocytes secretedboth
MMP-1 (29.8(cid:1)2.6 ng per mg protein after 48 h of incuba-
tion,Fig7)andMMP-13(48.1(cid:1)2.9ngpermgproteinafter
48 h of incubation; data not shown). In contrast, neither
collagenasecouldbedetectedinSZ95sebocytes(TableI).
Arachidonic acid significantly increased MMP-1 in HaCaT
keratinocytes in a time-dependent manner (Fig 7), but did
notinfluenceMMP-13(datanotshown).Isotretinoin(at10(cid:2)8
and 10(cid:2)7 M) did not influence the basal (Fig 7) or the
arachidonic acid-induced MMP-1 secretion in HaCaT ker-
atinocytes up to 48 h of incubation (data not shown). Is-
otretinoin appeared to reduce MMP-13 production in
HaCaTkeratinocytes,inatime-andconcentration-depend-
ent manner (14%–20% reduction), but this effect was not
statistically significant (data not shown).
Figure6
HaCaTkeratinocytesandSZ95sebocytessecretepromatrixmet-
alloproteinase pro(MMP)-2 and -9, which are upregulated by
arachidonic acid and downregulated by isotretinoin. Representa-
tivegelatinzymographiesfromthecellculturemediumofHaCaTker-
atinocytes (a) and SZ95 sebocytes (c), incubated for 24 h. Lanes 1:
untreated (control), lanes 2: arachidonic acid (AA; 10(cid:2)4 M), lanes 3:
isotretinoin (Iso; 10(cid:2)8 M), lanes 4: isotretinoin (10(cid:2)7 M), lanes 5:
arachidonic acid (10(cid:2)4 M)þisotretinoin (10(cid:2)8 M), lanes 6: AA (10(cid:2)4
M)þisotretinoin(10(cid:2)7M),lanes7:untreated(control),lanes8:N-ethyl-
maleimide(NEM,5mM),lanes9:Na EDTA(EDTA,20mM),andlanes
2
10:1,10-phenanthroline(Phen,4mM).ArrowsindicateproMMP-9(92
kDa),proMMP-2(72kDa),andMMP-2(64kDa).Quantitativeanalysis
of the gelatinolytic activity present in the culture media fromcell cul-
tures of HaCaT keratinocytes (b) and SZ95 sebocytes (d). Each bar
representsthemean(cid:1)SDfrom4zymograms.Statisticalsignificance:
(cid:1) (cid:1)(cid:1) (cid:1)(cid:1)(cid:1)
po0.05; po0.02, po0.01,ascomparedwithcontrol(untreated
cells)(),orwithAAalone[].Iso,isotretinoin;AA,(10(cid:2)4M).
2005 MATRIX METALLOPROTEINASESIN ACNEVULGARIS 7
SecretionofTIMP MeasurementofTIMPbyELISAindicat- (po0.05), and arachidonic acid-induced gene expression
ed the presence of TIMP-1 (16.7(cid:1)4.2 ng per mg protein was downregulated by isotretinoin by 33%. MMP-9 was
after 48h of incubation) and TIMP-2 (39.4(cid:1)5.5 ng per mg enhanced by arachidonic acid by 36% (po0.05), and both
protein after 48 h of incubation, data not shown) in HaCaT basal-andarachidonicacid-inducedgeneexpressionwere
keratinocytes, whereas neither TIMP could be detected in downregulated by isotretinoin by 42% and 49%, respec-
SZ95sebocytes(TableI,FigS1).Arachidonicacid(10(cid:2)4M), tively,(po0.05).TIMP-1wasenhancedbyarachidonicacid
significantly increased TIMP-1 in HaCaT keratinocytes in a by 60% (po0.05) but was not affected by isotretinoin. (b)
time-dependent manner (63% increase after 12 h to 132% SZ95 sebocytes: Arachidonic acid enhanced the gene ex-
increase after 48 h of incubation, po0.01), but did not in- pression of MMP-2 by 38% (po0.05), and isotretinoin
fluence TIMP-2 (data not shown). Isotretinoin (at 10(cid:2)8 and downregulated the arachidonic acid-induced gene expres-
10(cid:2)7 M) did not influence either basal TIMP or the sionby48%(po0.05).Similarly,arachidonicacidenhanced
arachidonicacid-inducedTIMP-1productioninHaCaTker- MMP-9 by 42% (po0.05), and isotretinoin downregulated
atinocytes up to 48 h of incubation (data not shown). the arachidonic acid-induced gene expression by 44%
(po0.05).
With the exception of MMP-13 (Fig 8, inset), however,
Gene expression of MMP and TIMP-1 Gene expression of
caution should be exercised in interpreting the changes in
MMP and TIMP-1 was investigated by RT-PCR analysis.
gene expression of other MMP and TIMP-1 in treated Ha-
MMP-1, -13, -2, -9, and TIMP-1 mRNA were expressed in
CaT keratinocytes and SZ95 sebocytes, as detected by
HaCaTkeratinocytesafter24hofincubation(Fig8,TableI),
semi-quantitativeRT-PCR,sincethemagnitudeofchanges
and MMP-2 and -9 in SZ95 sebocytes (Table I). Glyceral-
in expression relative to GAPDH were less than 2-fold,
dehyde-3-phosphate dehydrogenase gene (GAPDH) was
which is generally not considered to be significant, despite
used as the internal standard. Quantification of chemilumi-
the statistically significant differences in the relative chemi-
nescencewasperformedusingacomputer-assistedimage
luminescence values obtained from the ethidium bromide-
analysis program. The ratio of chemiluminescence of each
stained gels.
parameter measured to GAPDH revealed the following: (a)
HaCaT keratinocytes: The gene expression of MMP-1 was
enhanced by arachidonic acid (10(cid:2)4 M) by 42%, (po0.05),
but was not affected by isotretinoin (10(cid:2)7 M). MMP-13 (Fig Discussion
8, inset) was enhanced by arachidonic acid by 37% and
In this study, we investigated the involvement of MMP and
both basal- and arachidonic acid-induced gene expres-
TIMP in sebum from facial acne lesions, and the effect of
sions were downregulated by isotretinoin by approximately
isotretinoin treatment on these molecules associated with
50%(po0.05).ArachidonicacidenhancedMMP-2by37%
inflammatory matrix responses. Gelatin zymography and
western blotanalysisindicatedthepresenceofgelatinases
in sebum, attributed mainly to proMMP-9, whereas ELISA
and western blot analysis revealed the presence of colla-
genases MMP-1, mainly as proMMP-1 and MMP-13, both
inlatentandactiveforms.TheexpressionofTIMP-1and-2
was also demonstrated using ELISA, and their presence
may account for the existence of MMP mainly in the latent
form.
The predominant types of cells that could be identified
under light microscopy of sebum specimens were keratin-
ocytes, P. acnes, and Staphylococcus epidermidis. Light
microscopy of sebum samples revealed that keratinocytes
andbacteriacountdecreasedwithprogressionoftreatment
with isotretinoin.
The cell source of MMP and TIMP in sebum that we
Figure8 observed cannot be attributed to P. acnes or other micro-
HaCaTkeratinocytesexpressmRNAformatrixmetalloproteinase
organisms found in the pilosebaceous unit, such as
(MMP) and tissue inhibitor of matrix metalloproteinase (TIMP)-1,
and isotretinoin inhibits MMP-9 and MMP-13 de novo and Pityrosporum ovale, Propionibacterium granulosum, and
arachidonic acid-induced mRNA. Quantitative analysis of chemilu- S. epidermidis, since these bacteria do not comprise a
minescenceofRT-PCRanalysesofHaCaTkeratinocytesincubatedin
source of MMP activity. Therefore, the cell origin of MMP
the absence or in the presence of isotretinoin (10(cid:2)7 M, 24 h) using
primersforMMP-1,MMP-9,MMP-13,andTIMP-1,asdescribedunder andTIMPinsebumislikelytobekeratinocytes.Sebocytes
‘‘Materials and Methods’’. Each bar represents the mean(cid:1)SD from may also be responsible for MMP and TIMP expression in
(cid:1)
four assays. Statistical significance: po0.05; AA, arachidonic acid, sebum,since the latter is a holocrine product of sebocytes
10(cid:2)4 M; Iso, isotretinoin, 10(cid:2)7 M. Inset: Representative analysis of
(Zouboulis et al, 2003). The possibility that keratinocytes
chemiluminescenceofRT-PCRanalysesofHaCaTkeratinocytesincu-
batedintheabsenceorinthepresenceofisotretinoin(10(cid:2)7M,24h) and sebocytes may comprise the cell source for MMP and
usingprimersforMMP-13.GAPDHofexpectedsize263bpwasused TIMP in sebum is supported by our findings that HaCaT
asaninternalstandard.Theexpectedsizeofthereactionproductfor keratinocytes in culture express and secrete proMMP-2,
MMP-13isindicatedbythearrow.Lane1,DNAladder(100bp);lane2,
untreated;lane3,AA(10(cid:2)4M);lane4,isotretinoin(10(cid:2)7M);andlane5, proMMP-9,MMP-1,MMP-13,andTIMP-1andTIMP-2,and
AA(10(cid:2)4M)andisotretinoin(10(cid:2)7M). that SZ95 sebocytes in culture express and secrete pro-
8 PAPAKONSTANTINOU ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
MMP-2 and -9, which was also confirmed by cDNA micro- The effects of isotretinoin in sebum MMP and TIMP fol-
array analysis. Sebocytes may be a source of other MMP, lowing topical or systemic treatment of acne patients cor-
which may also contribute to the pathogenesis of relatewiththeeffectsinducedbyisotretinoininarachidonic
acne. These findings are in agreement with reports that acid-treatedHaCaTkeratinocytes, whichmaybetterrepre-
human keratinocytes of normal or pathological origin pro- sent follicular keratinocytes of acne patients than HaCaT
duce or express latent and active gelatinases and TIMP keratinocytesnotunderinflammatorychallenge.Isotretinoin
(Baumann et al, 2000; Fleischmajer et al, 2000; Kobayashi significantly inhibited the arachidonic acid-induced proM-
et al, 2000). MP-2 and -9 and MMP-13 secretion by HaCaT keratin-
ProMMP-2 was not detected in sebum, even though ocytes, although the latter was not statistically significant,
both HaCaT keratinocytes and SZ95 sebocytes produced as well as the arachidonic acid-induced mRNA of MMP-9
proMMP-2. This may be attributed to the fact that cells in and -13. In SZ95 sebocytes, isotretinoin also inhibited the
culture are not confronted with proteases at the level that arachidonic acid-induced increase of proMMP-2 and -9
inflammatory human tissue does. In the latter case, the secretion, and the arachidonic acid-induced expression of
presence of proteases may lead to deactivation, inhibition, mRNA for MMP-9. The above-described effect of is-
or destruction of MMP-2 in the diseased human follicle otretinoin on MMP production by keratinocytes and se-
in vivo. Even though the results from HaCaT keratinocytes bocytes is in good agreement with studies on the effect of
and SZ95 sebocytes in culture are of considerable value, RAonvariouscellculturesortissues,whichdemonstrated:
caution should be exercised in extrapolating the results (a)thatall-trans-RAinhibitedtheactivityofMMP-2and-9in
from cellcultures to the invivo situation, sinceHaCaTcells human arterial smooth muscle cells (Axel et al, 2001), the
are not a model for follicular keratinocytes and differences activity of MMP-9 in human bronchoalveolar lavage cells
may exist in MMP expression between normal and HaCaT (Frankenbergeretal,2001),theactivityofMMP-2and-9in
keratinocytes. Thus, the possibility that the cell sources of a rabbit model of vein bypass grafting (Leville et al, 2000),
sebum MMP and TIMP are follicular keratinocytes remains and pro- and active MMP-9 protein in myelogenous le-
to be verified by using tissues of sebaceous follicle and ukemia cell lines (Devy et al, 2002); (b) that retinol and all-
glands of acne patients. trans-RA decreased 98–96, 72–68, and 46–45 kDa gelatin-
The prospect that other type of cells associated with olytic activity in capillary endothelial cells (Braunhut and
MMP and TIMP production, such as skin fibroblasts or in- Moses, 1994); (c) that retinoids reduced the activation of
filtrating macrophages in the inflammatory acne lesions, MMP in fibroblasts (Overall, 1995), and (d) that locally pro-
may comprise the cell source of MMP and TIMP in sebum duced all-trans-RA suppresses MMP during endometrial
should also be considered. In this respect, it has been re- differentiation(Osteenetal,2002).Similarly,thelackofany
ported that retinoids reduce the activation of MMP in fib- effect of isotretinoin on TIMP secretion in sebum or in cell
roblasts (Overall, 1995) and that tretinoin downregulates culturesisinagreementwithreportsindicatingthatoralall-
MMP-9 mRNA and protein in alveolar macrophages from trans-RA did not affect TIMP-1 in a rabbit model of vein
patientswithchronicobstructivepulmonarydisease(Frank- bypass grafting (Leville et al, 2000), and that it did not in-
enbergeretal,2001)andemphysema(Maoetal,2003),and fluence expression of TIMP-1 in human primary melanoma
that it reverses upregulation of MMP-13 in human keloid- (Jacob et al, 1998).
derived fibroblasts (Uchida et al, 2003). The inhibition of proMMP-9 following treatment with is-
The pro-inflammatory essential fatty acid arachidonic otretinoin is not due to a direct inhibitory effect of the drug
acid significantly induced proMMP-2 and -9 in both cell on the enzyme, since isotretinoin, up to 1 mM, did not ex-
typesinvestigated,aswellasMMP-1andTIMP-1inHaCaT hibit any direct effect on proMMP-9 isolated from sebum
keratinocytes. The association between arachidonic acid specimens of facial acne lesions or on commercially avail-
and MMP and TIMP expression has also been reported for able gelatinases. The decrease in proMMP-9 and MMP-13
several cell types and tissues, as e.g., in the human athe- that we observed following treatment with isotretinoin may
rosclerotic plaques (Cipollone et al, 2003), in synoviocytes beattributedtothereductionoftheexpressionrateofthese
(Burger et al, 2003), in renal tubular cells (Cussac et al, MMP. This is in agreement with our observation that is-
2002), and in calvaria and bone marrow cells (Choi et al, otretinoin downregulated MMP-9 and -13 mRNA in HaCaT
2003). keratinocytes and MMP-9 in SZ95 sebocytes. It is also
Topical or systemic treatment of acne patients with is- supported by reports that RA reduces the mRNA expres-
otretinoin did not affect MMP-1 or TIMP but resulted in re- sionofMMP-2and/or-9inhumanarterial(Axeletal,2001)
duced activity of MMP-9 and secretion of MMP-13 in and rat aortic (Neuville et al, 1999) smooth muscle cells, in
sebum, an effect that was enhanced with the duration of tumor (Tsang and Crowe, 2001) and leukemia (Devy et al,
treatment and in parallel to the improvement of the clinical 2002) cell lines, and in a rabbit model of vein bypass graft-
picture. The effects of per os or topical administration of ing (Leville et al, 2000), whereas tretinoin reverses upreg-
isotretinoin on proMMP-9 of sebum samples that we ob- ulation of MMP-13 in human keloid-derived fibroblasts
served are in good agreement with reports that per os ad- (Uchida et al, 2003).
ministration of all-trans-RA to patients with emphysema ThepresenceofMMP-9activityandprotein,andofcol-
reduced plasma MMP-9 protein and activity, while having lagenases MMP-1 and -13 and TIMP protein of epithelial
little effect on TIMP-1 levels (Mao et al, 2003), and that origininacnesebum,andtheisotretinoin-induceddecrease
topical administration of retinol in healthy humans reduced of proMMP-9 and MMP-13 in parallel to the clinical im-
MMP-9 activity in punch biopsies of human buttock skin provementofthelesionsindicatethattheseMMPandTIMP
(Varani et al, 2000). may be involved in the pathophysiology of acne lesions,
2005 MATRIX METALLOPROTEINASESIN ACNEVULGARIS 9
possiblybycontributingtoabnormalhyperproliferation,and penicillin/streptomycin (all from Biochrom) at 5% CO2 and 371C.
degradation and remodeling of extracellular matrix struc- Cells werecultured for 2 d in the presenceor absence of arachi-
donicacid(Sigma-Aldrich,Deisenhofen,Germany),anddissolved
tures, such as the basement membrane of the acroinfundi-
inethanoltogiveafinalarachidonicacidconcentrationof10(cid:2)4M.
bulumof the sebaceous follicle and the sebaceous glands.
The final concentration of ethanol in medium without and with
Although the precise functional role of MMP and TIMP in
arachidonicacidwas0.1%.Subsequently,cellsweretreatedwith
acne pathology remains to be clarified, it appears that the isotretinoin(Sigma-Aldrich)dissolvedindimethylsulfoxidetogive
isotretinoin-induced reduction in proMMP-9 and MMP-13, final isotretinoin concentrations of 10(cid:2)8 and 10(cid:2)7 M. The final
via mechanisms that do not affect TIMP, may contribute to concentration of dimethyl sulfoxide in medium without and with
the therapeutic effects of this agent in acne. isotretinoin was 0.2%. Isotretinoin was handled under dimmed
yellowlight.Culturesupernatantswerecollectedafter12,24,and
48 h of treatment with isotretinoin or DMSO into sealed plastic
tubesandfrozenat(cid:2)401Cuntilfurtherevaluation.Cellswerealso
Materials and Methods
collectedseparatelyafter48hoftreatmentandstoredundersim-
ilarconditions.
Patients Lesions in acne vulgaris patients ranged from clinically
non-inflammatory microcomedones, closed or open comedones,
toinflammatorypapules,pustules,andcysts,intermingledtovar- Gelatinzymography Gelatinzymographyanalysiswasperformed
in sebum specimens and medium of cell cultures, collected as
ious extents. The criteria of the Global Alliance to Improve Out-
described above. In addition, gelatin zymography was performed
comesinAcnewereusedtoclassifyacneintomild,moderate.and
severe (Gollnicketal, 2003). Fifty-nine female patients with acne following invitroexperiments to test the effects of the agent on
gelatinasespresentinsebumaswellasoncommerciallyavailable
vulgaris,aged17.8(cid:1)1.7y(mean(cid:1)SD)weretreatedwithtopical
gelatinases.Sebumsamplesweresuspendedin1mLddH Oand
(0.05% once daily, 36 patients with mild and moderate papulo- 2
subjected to ultrasonication in a Clifton Ultrasonic bath (Nickel-
pustularacne)orsystemic(1mgperkgperdperos,23patients
with moderate nodular or severe acne) administration of is- Electro,Weston-Super-Mare,NorthSomerset,UK)(3(cid:3)5min)and
precipitation from saturation with 25%–50% (NH ) SO at 41C
otretinoin for 3–4 mo, after providing their written consent. The 42 4
(Karakiulakis etal, 1988). For cell cultures, aliquots of the supe-
medical history of all patients was free from recent microbial in-
rnatants were diluted to contain the same amount of protein per
fectionsoranyotherdisordersoftheskin.Patientshadnotbeen
treated with retinoids in the past and were not under any other mL. The gelatinolytic activity of MMP was determined by gelatin
zymographyanalysisusingsodiumdodecylsulfate-polyacrylamide
medicationforatleast3mopriortotheinitiationofthetreatmentin
gel electrophoresis(SDS-PAGE) under denaturing but non-reduc-
thisstudy.Allpatientswerethoroughlyinformedabouttheadverse
effectsofisotretinointreatmentandreceivedother,butnohormo- ingconditions(Karakiulakisetal,1997).Molecularsizesofbands
displaying enzymatic activity were estimated in comparison with
nal, contraception during and 3 mo after discontinuation of the
purifiedproMMP-2(72.0kDa),activeMMP-2(64.0kDa),proMMP-
treatment. Participants gave their written informed consent. The
9 (92.0 kDa) and active MMP-9 (68.0 kDa) (Anawa Trading, Wan-
medical ethical committee of Aristotle University of Thessaloniki
hasapprovedallthedescribedstudies.Thestudywasconducted gen). The pre-stained standard protein molecular weight markers
used were: phosphorylase b (97.4 kDa), bovine serum albumin
accordingtotheDeclarationofHelsinkiPrinciples.
(66.2 kDa), L-glutamic dehydrogenase (55.0 kDa), ovalbumin
Sebumsamples Sebumfromfaciallesions,suchascomedones, (42.7 kDa) and aldolase (40.0 kDa) (all from Promega, Madison,
papules, and pustules, were collected under standardized condi- WI).Gelatinolyticactivitywasquantifiedusingacomputer-assisted
tions from individual acne lesions. Sebum collection was per- imageanalysisprogram(1DImageAnalysisSoftware,version3.0
formedusingsterile,blunt,plastic‘‘spatulas’’torupturethelesion ofKodak DigitalScience, Eastman Kodak, Rochester,New York).
andgentlysqueezeoutthesebum.Sebumwascollectedatthetip The nature of the proteolytic bands was further characterized by
of the spatula and transferred into sterile 1 mL Eppendorf plastic including specific protease inhibitors: Na2EDTA (20 mM), 1,10-
vials.Sampleswerepooledinoneplasticvialperindividualpatient, phenanthroline (4 mM), orN-ethyl-maleimide (5 mM) (all obtained
per time point, and stored at (cid:2)701C until use. Sebum sampling fromSigma-Aldrich)intheenzymeincubationbuffer.Theeffectof
wasperformedatthreetimepoints:priortotreatment(0dsample), isotretinoinon0dsebumsamples(correspondingto5mgprotein)
31(cid:1)3d(30dsample),and63(cid:1)4d(60dsample)aftertreatment orpurifiedlatentMMP-2and-9(1ngprotein)wasstudiedbypre-
withisotretinoin.Aliquotsofsebumsamples,dilutedtocontainthe incubating samples withisotretinoin (0–1mM)for 30 minat room
sameamountofproteinpermL,wereanalyzedinablindedfashion temperature, followed by gelatin zymography. SDS was then re-
forthepresenceorabsenceofkeratinocytesand/orbacteria.Light movedfromthegelsbyequilibrating(2(cid:3)30min)in2.5%(vol/vol)
microscopyof sebumsamples was performed using an Olympus Triton X-100 (Sigma-Aldrich) and gels were incubated in enzyme
BX50 microscope (Tokyo, Japan). Bacterial cultures of sebum buffer (50 mM Tris-HCI, pH 7.3, containing 200 mM NaCI, 5 mM
specimensobtainedbeforeandduringtreatmentwereperformed CaCl2and0.1%TritonX-100),inthepresenceofisotretinoin(0–1
under strict aseptic techniques in Columbia blood agar and an- mM)for18h,at371C.
aerobe blood agar. Bacteria were identified on the basis of mor-
phological and biochemical characteristics employing the VITEK Western blot analysis for determination of MMP Sebum sam-
system (bioMerieux, Marcy l’Etoile, France). Identification was ples,containingthesameamountoftotalprotein,wereenrichedin
confirmedbytestingforvariouspropertiesusingtestsaccordingto MMPbyprecipitationwith(NH4)2SO4(60%saturation)(Karakiula-
Murrayetal(1999). kis etal, 1988); the resultant precipitates were dissolved in La-
emmlisamplebuffercontaining5%b-mercaptoethanol,boiledfor
Cultures and treatment of HaCaT keratinocytes and SZ95 se- 5min,andthensubjectedtoSDS-PAGE(Laemmli,1970)on10%
bocytes Spontaneouslyimmortalized,nontumorigenichumanHa- polyacrylamidegels.Afterelectrophoresis,theseparatedproteins
CaTkeratinocytes(Boukampetal,1988)(generouslyprovidedby wereelectro-transferredontonitrocellulosemembranesaccording
Prof.N.E.FusenigandDrD.Breitkreuz,GermanCancerResearch tothemethodofTowbinetal(1979).Thefreebindingsitesonthe
Center,Heidelberg,Germany)andimmortalizedhumanfacialSZ95 nitrocellulose membranes were blocked with 5% skim milk in 20
sebocytes(Zouboulisetal,1999)wereseededataconcentration mMTris-HCl,pH7.4/150mMNaClbuffer(TBS),containing0.05%
of 3(cid:3)104 cells per well in 24-well culture plates (Nunc, Wiesb- Tween-20(TBS-T),atroomtemperaturefor1h.Afterthree10min
aden,Germany)andweremaintainedinSebomedmedium(Bioch- washes with TBS-T, the membranes were incubated with rabbit
rom, Berlin, Germany) containing 10% (vol/vol) fetal calf serum, antiserum produced against human MMP-2 or MMP-9, (a gener-
5 ng per mL recombinant human epithelial growth factor, and ous gift from Dr. P. Koolwijk, Gaubius Lab. TNO-PG, the Nether-
10 PAPAKONSTANTINOU ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
lands)(Hanemaaijeretal,1998),oraffinitypurifiedanti-rabbitpo- plification specific for GAPDH (MWG-Biotech AG) (263 bp, an-
lyclonal AB806 for MMP-1 and AB8114 for MMP-13 (Chemicon nealingat611C,30cycles)wasusedtoestimatetheefficiencyof
International, Temecula, California), at dilution 1:1000, in TBS-T, the reverse transcription reaction for MMP and TIMP-1. Five mi-
containing1%skimmilk,at41Cfor20h.Afterwashingthreetimes croliters of each PCR reaction mixture were analyzed in a 2%
withTBS-T,theywereincubatedwithperoxidase-conjugatedgoat agarose gel, using a 100 bp DNA ladder (Invitrogen, Life
anti-rabbit IgG at dilution 1:4000 in TBS-T, containing 1% skim Technologies, Carlsbad, California). Visualization of DNA bands
milk, at room temperature for 2 h. Then, the membranes were was achieved with UV illumination of ethidium bromide-stained
washedwithTBS-Tthreetimes,oncewithTBSandtheimmuno- gels.Quantificationofchemiluminescencewasperformedusinga
reacted proteins were detected by the enhanced chemilumines- computer-assisted image analysis program (1D Image Analysis
cence method, according to the manufacturer’s instructions Software,version3.0ofKodakDigitalScienceEastmanKodak).
(Pierce, Rockford, Illinois). For negative controls, nitrocellulose
membranes were subsequently treated with stripping solution Analysis of MMP and TIMP expression by cDNA microar-
(Chemicon). After stripping, membranes were re-blocked, as de- rays SZ95 sebocytes were maintained under the conditions de-
scribedabovewithoutarachidonicacidandisotretinoinfor120h.
scribed above, and incubated with rabbit anti-ovalbumin IgG
Subsequently, RNA extraction was performed using the RNeasy
(Chemicon)(insteadofthespecificrabbitantibodiesforMMP),at
dilution1:1000,inTBS-T,containing1%skimmilk,at41Cfor20h. Midi kit (Qiagen) according to the manufacturer’s protocol. RNA
Subsequent steps were as described above. When the non- wasphotometricallymeasuredinaPharmaciaGeneQuantIIspec-
specific antibody was used, no immunoreactive bands could be trophotometer(Freiburg,Germany)andstoredat(cid:2)801Cuntiluse.
AmplifiedRNA(aRNA)wasgeneratedfrom3mgofDNase1-treat-
detectedforanyMMPexamined.Quantificationofchemilumines-
cence was performed using a computer-assisted image analysis edtotalRNAusingtheMegaScriptT7HighYieldTranscriptionkit
(Ambion, Austin, Texas). RNA purity, integrityand concentrations
program(1DImageAnalysisSoftware,version3.0ofKodakDigital
were evaluated on the Agilent 2100 bioanalyzer. Cy3- and Cy5-
Science).Molecularsizewasestimatedbycomparisonprestained
labelled cDNA was reverse transcribed from 3 mg of aRNA per
standard proteins: phosphorylase b (97.4 kDa), bovine serum al-
bumin (66.2 kDa), ovalbumin (45.0 kDa) and soybean trypsin in- reaction. All labeling reactions used the Cyscribe First-Strand
cDNA labeling Kit (Amersham Pharmacia). Purification of labeled
hibitor (21.5 kDa) (all from Promega, Madison, Wisconsin), which
cDNA was carried out using Microcon YM-30 columns (Millipore,
wereelectrophoresedunderreducingconditions.
Billerica,Massachusetts).Fourreplicatedhybridizationsconsisting
ofduplicateddyeswapswerecarriedoutontheHumanENSEMBL
ELISAfordeterminationofcollagenasesandTIMP MMP-1and
chipusingprotocolsdescribedpreviously(Adjayeetal,2004).The
-13 were determined in aliquots of sebum samples or the supe-
cDNA microarray consists of 15,500 non-redundant, fully se-
rnatantsofcellculturesbyELISA.SampleswereenrichedinMMP
quenced,annotatedhumancDNA(HumanEnsemblsetRZPD1.1)
byprecipitationwith(NH ) SO (60%saturation)(Karakiulakisetal,
42 4 spottedinduplicateonsuperAmine-coatedglassslides(Telechem,
1988) and assayed for MMP-1 and -13 as previously described
Sunnyvale, California) with each slide containing positive/internal
(Papakonstantinouetal,2003),usinganti-MMP-1(1mgpermL,Ab
controls(b-actin, HPRT), and a selection of Arabidopsis cDNAas
806,Chemicon)oranti-MMP-13(1mgpermL,Ab8114,Chemicon)
negative controls. Slides were scanned using the Affymetrix 428
polyclonalantibodiesinPBS-Tandperoxidase-conjugatedsecond
ArrayScanner(SantaClara,California).
antibody (goat anti-rabbit IgG, Chemicon). The concentration of
collagenases was estimated in pg collagenase protein per mg of
DataanalysisofcDNAmicroarrays Imageanalysiswascarried
totalproteinpersebumsample,usingreferencesamplesofMMP-
out with the AIDA Array Matrix software (Raytest, Straubenhardt,
1 and -13 from human periprosthetic tissue of loose hip endo-
Germany).Intotal,11replicateexperimentswerecarriedout.Data
prostheses(Syggelosetal,2001).TIMP-1and-2weremeasuredin
were normalized as described in Herwig etal(2001). In order to
aliquots of sebum samples or the supernatants of cell cultures,
judgewhetheragivengenewasexpressedinSZ95sebocytes,for
using ELISA systems (Biotrak RPN 2611 for TIMP-1 and Biotrak
each experiment, the signal of the gene was compared with a
RPN2618forTIMP-2;AmershamPharmacia, Freiburg,Germany)
reference distribution derived from 3626 negative control signals
thatrecognizetotalhumanTIMP-1or-2,bothfreeandthatcom-
by computing the proportion of negative spots having a smaller
plexedwithMMP.
signalthanthegeneofinterest.Theaverageproportionacrossall
replicateswasdefinedasthe‘‘expressionprobability’’forthegene
RT-PCR for expression of MMP and TIMP-1 Isolation of RNA andwasusedasanindicatorfortheexpressionstrengthinSZ95
was performed by the RNeasy spin mini kit (Qiagen, Hilden, Ger-
sebocytes.Visualinspectionofhybridizationimagesindicatesthat
many),accordingtothemanufacturer’sinstructions.Geneexpres- ahighprobability(p40.95)correspondstovisiblesignals,whereas
sionofMMPandTIMP-1wasascertainedbysemi-quantitativeRT- spots with po0.9 correspond to absent signals. Probabilities in
PCRanalysisaccordingtoPapakonstantinouetal(2004),adjusted betweencorrespondtoweaklyexpressedsignals.
asfollows:totalRNA(50ngforMMP-1,400ngforMMP-13,200
ngforMMP-2,200ngforMMP-9,20ngforTIMP-1,and10ngfor Proteinmeasurement Proteincontentwasdeterminedinaliquots
GAPDH) was added to each RTreactioncontaining one-step RT- ofsebumspecimens,aliquotsofcellculturesupernatantsandcells
PCRmix(RobusI,Finnzymes,Espoo,Finland)andtheappropriate with the standard Bradford assay (Bio-Rad, Glattbrugg, Switzer-
PCRprimers(15pmole)in50mLtotalvolume.Reversetranscrip- land) using bovine serum albumin (Sigma-Aldrich Chemie, Stein-
tionwascarriedoutfor30minat501C,followedbya2minstepat heim, Germany) as standard. All data presented were normalized
941C. PCR was then performed on a PTC-100 programmable perproteincontentforeachsebumorcellculturespecimen.
ThermalController(MJResearch,Waltham, Massachusetts), pro-
grammed for several cycles of 1 min at 941C, 1 min at optimal Statisticalanalysis Whererelevant,dataarepresentedasmean
annealing temperature, and 1 min at 721C, followed by a 10 min (cid:1)SD. Differences between means were evaluated by anal-
stepofextraextensionat721C.TheamountoftotalRNAused,as ysisofvariance.po0.05wasconsideredstatisticallysignificant.
wellasthenumberofPCRcycles,wasadjustedsothateachPCR
amplificationwasinthelinearrange.PrimersforMMP-1(428bp,
annealing at 551C, 30 cycles, Konttinenetal, 1999), MMP-2 (605
bp,annealingat581C,40cycles,Giambernardietal,1998),MMP-
We acknowledge Prof. N. E. Fusenig and Dr D. Breitkreuz, German
9(519bp,annealingat551C,30cycles,Mooreetal,2000),MMP- CancerResearchCenter,Heidelberg,Germanyforkindlyprovidingthe
13(517bp,annealingat531C,35cycles,Mooreetal,2000),and HaCaT keratinocyte line, and Dr P. Koolwijk, Gaubius Lab. TNO-PG,
TIMP-1(534bp,annealing at601C,30cycles,Mooreetal,2000) The Netherlands, for kindly providing rabbit antiserum produced
wereobtainedfromMWG-BiotechAG(Ebersberg,Germany).Am- againsthumanMMP-2orMMP-9.Thisworkwassupportedbygrants